Classification is one of several common tasks undertaken by machine learning. This paper provides an overview of some machine learning architectures (Support Vector Machines, Multi-Layer Perceptron, Convolutional Neural Networks) and evaluates their suitability for classification within the domain of computer vision. Then, we proceed to evaluate the performance of such architectures under varying hyperparameters and other conditions, such as neural network shape, regularization, and augmented data, to inquire into methods for optimizing model performance in respect of image classification.
Refer to the report in this repository. This repository also contains supporting notebooks. This was a final deliverable for the UdeM course introduction to machine learning https://admission.umontreal.ca/cours-et-horaires/cours/ift-6390/
- Alexander Peplowski
- Azfar Khoja
- Daniel Wang